View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by I-Revues DOES NEST PREDATION AFFECT NEST-SITE SELECTION IN LARKS ? Miguel YANES and Juan J. ONATE• INTRODUCTION Nest predation is one of the main factors determining avian reproductive strategies (Lack, 1968 ; Ricklefs, 1969). One adaptive response to it is nest-site selection as slight differences in the area around the nest may result in different nest predation rates. Such is the case with, for example, habitat structure around the nest (Martin & Roper, 1988 ; Clark & Nudds, 1991 ; Li & Martin, 1991 ; Mankin & Warner, 1992 ; Riley et al. , 1992) or distance to the ecosystem edge (Rodenhouse & Best, 1983 ; Small & Hunter, 1988). Ground nesting birds experience high nest predation, especially those that breed in shrub and grassland habitats (Martin, 1993 ; Yanes & Suarez, 1995). In holarctic shrubsteppes, nest predation rates reach values of over 60 %, although they vary a lot with area and year (Rotenberry & Wiens, 1989 ; Suarez & Manrique, 1992). Changes in management may even increase rates to values of around 90 % (Suarez et al. , 1993 ). In this paper we analyze the effect of predation on lark' s nest habitat selection, in an area of very high nest predation, in order to test the hypothesis that nest predation is a selective pressure shaping nest-site selection in order to reduce nest predation risk. • STUDY AREA AND METHODS STUDY AREA The study was carried out in the Ornithological Reserve of « Las Amola­ deras », Almerfa, SE Spain (36°50' N, 2°25' W, 0-50 rn asl) ; a 850 hectare shrubsteppe with a semi-arid Mediterranean climate (mean annual rainfall of 250-300 mm). Most of it is a plain with thin vegetation consisting of chamephytes such as Te ucrium belion, Helychrysumstoechas, Thymelaea hirsuta and Haloxylon articulatum, and, to a lesser degree, grasses as Stipa tenacissima and Ammophila arenaria. However, there is also an extensive abandoned Agave Agave spp. plantation and stands of a larger bush called « Azufaifo » Ziziphus lotus grow locally. * Dpto. Ecologfa, Facultad de Ciencias. Universidad Aut6noma. E-28049, Madrid, Spain. Rev. Eco/. (Terre Vie), vol. 51, 1996. - 259 - The reserve is partialiy bordered by two large « ramblas » (dry rivers) and crossed by other smalier ones. These ramblas, Agave plantations and azufaifo stands contain ali the fox dens in the study area (Yanes et al., in press). Red Foxes Vu lpes vulpes, together with ferai Dogs Canis fa miliaris, are the main nest predators in the study area (Yanes & Suarez, in press). Others are Bastard Snake Malpolon monspessulanus, Horseshoe Snake Coluber hipocrepis, Ladder Snake Elaphe scalaris, Eyed Lizard Lacerta lepida, Hedgehog Erinaceus europaeus and Great Grey Shrike Lanius excubitor. The bird community is very poor in species, larks being the dominant ones (Telieria et al. , 1988). Thekla Lark Galerida theklae and Lesser Short-toed Lark Calandrella rufescens are the only species that are relatively numerous. Their high nest predation rates do not appear to be affected by researcher visits (Westmore­ land & Best, 1985 ; Major, 1990), because i) in other years no relationship was observed between monitoring frequency and nest predation (Suarez & Manrique, 1992), and ii) no differences were noted in a series of experiments with artificial nests monitored daily and unmonitored ones (Suarez et al. , 1993). NEST-SITE HABITAT MEASUREMENTS Habitat features of 150 nests were measured (75 of Thekla Lark, 75 of Lesser Short-toed Lark) and 75 random sites. For each species 60 nests were predated and 15 successful. The nests were found using random field itineraries and visual observation of adult behaviour over three years ( 1992-94). The random points were taken in the same years, at 60 rn intervals along random linear tracks. Habitat characteristics were estimated in a circle centred on the nest or the random point during the nesting periods. Two circle sizes were considered : a smali 1.5 rn radius one, in the immediate vicinity of the nest, and a large 15 rn radius territorial one. Using the method proposed by Prodon (1976), we estimated on both scales the overali area covered by : shrubs < 20 cm high (variable S20 on smali scale and L20 on large scale), between 20-40 cm (S24 and L24), > 40 cm (S40 and L40), grasses (SGR and LGR), open ground (SBG and LBG) and rock (SRO and LRO). The chosen height of the shrubs for the variables corresponds to the most frequent sizes in the area. On the large scale we also considered two other 2 variables : distance to a « clearing » of bare ground or rocky substrate of over 6 m (CLE) and the distance to the shrubsteppe edge (EDG). Edge was taken to mean the boundary of the ramblas, the Agave plantations and the high azufaifo stands. STATISTICAL ANALYSES To summarize habitat characteristics, the variables were reduced to two orthogonal axes using principal component analysis (PCA). Two series of PCAs were carried out to discover the possible existence of (i) nest-site habitat selection and (ii) habitat differences in successful and predated nests. First two PCAs, one on each scale, smali and large, including ali nests (both successful and predated) of both species and random points were carried out. Secondly, another two PCAs for each species, one on each scale but differentiating successful and predated nests, were carried out. The differences in the principal component scores between these groups were analyzed using an analysis of variance (ANOVA). To compare the scores of the two species one with another and of one with the random sites, - 260 - a Tukey HSD multiple comparison was carried out. The variables were trans­ formed prior to the PCA : the arcsine transformation was used for variables expressed as percentages, and the logarithmic transfonnation was applied to the remaining ones. Moreover, we made individual comparisons of the untransformed habitat variables for species vs random sites and for successful vs predated nests, using the Mann-Whitney U test. The SYSTAT statistical package was used for ali statistical analyses (Systat, 1992). RESULTS NEST-SITE SELECTION The first small scale PCA axis orders the nests and random points along a gradient from places dominated by herbs to areas covered mostly with rock and low shrubs. The PCA scores for the nests and random points on this axis differ according to group (ANOVA, F ratio = 7.377 ; p < 0.01) ; there are significant differences between Lesser Short-toed Lark nests and random sites, but there are no differences between species neither between random sites and Thekla Lark nests, althoug in this last case the probability is near to significance level (Table 1). The second axis orders the vegetation along a complexity gradient from areas with high shrubs to areas with more bare ground. In this case also the differences are significant (F ratio = 7.107 ; p < 0.01 ), due to the segregation of the nests of the different species and of Lesser Short-toed Lark compared with the contrais (Table 1). These first two axes together accounted for 53.3 % (30 . 6 and 22.7 %, respectively) of the variance of the original six variables. TABLE 1 Probability values of the differences between PCA scores according to the Tu key HSD multiple comparison. Thekla Lark Thekla Lark L. Short-toed Lark VS vs vs L. Short-toed Lark Random sites Random sites Small scale Axis I 0.240 0.069 0.000 Axis II 0.001 0.779 0.011 Large scale Axis I 0.556 0.809 0.005 Axis II 0.000 0.017 0.001 The configuration of the large scale PCA axes is similar to that of the previous ones. On the first axis the variables of greatest weight are the low matorral and least distance to the clearing in relation to pasture caver. However, there are no - 26 1 SMALL SCALE 3 • • 0 • o 2 0 c;o • 0 • C\1 & .· � 1 • a: .a. • � . u 0 LE -1 -2 -2 -1 0 1 2 FACTOR 1 lARGE SCALE 2 • 0 C\1 0 a: 0 D � u -1 LE • • • -2 • 0 • -3 • -2 -1 0 1 2 3 FACTOR 1 Figure 1. - Ordering on both scales according to the PCA scores for the nests of the two larks and random points. Circles : Thekla Lark ; Squares : Lesser Short-toed Lark ; Points : Random sites. - 262 - differences between groups (F ratio = 1.395 ; p > 0.05), whether they are nests or random points (Tab. 1). The second axis ranks the points from areas covered in matorral of medium or large size to those with a greater area of open land. Unlike the first there are important differences between groups (F ratio = 27.289 ; p < 0.01) ; not only are the species clearly separated one from another but also from the random sites (Table 1). The variance accounted for by the two axes of the eight initial variables is 51.8 % (28.1 and 23.7 %, respectively). The arrangement of both species' nests and random points in the space formed by PCA axes, for the small scale as for the large one, is shown in shown in Figure 1. Considering the variables individually, Lesser Short-toed Lark is the species showing the most marked nest-site selection : ali the variables on the small scale and over half on the large scale differ significantly from the random sites (Table Il). Thekla Lark is the most eclectic when choosing a nest-site, tending to select only sites far from the shrubsteppe edge, with greater rock cover in the immediate vicinity of the nest and with few high shrubs on both scales (Table Il). The greatest differences between species occur on the large scale (Table Il). TABLE II Mean ± sd of considered habitat variables in random sites and nests.